Gait & Posture 42 (2015) 158–164
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Effects of temporal constraints on medio-lateral stability when negotiating obstacles Wataru Nakano a,*, Takashi Fukaya a, Yoshihide Kanai b, Kazunori Akizuki c, Yukari Ohashi b a
Department of Physical Therapy, Tsukuba International University, 6-8-33 Manabe, Tsuchiura, Ibaraki, Japan Department of Physical Therapy, Ibaraki Prefectural University of Health Sciences, 4669-2 Ami, Ami, Ibaraki, Japan c Department of Physical Therapy, Mejiro University, 320 Ukiya, Iwatsuki, Saitama, Japan b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 28 July 2014 Received in revised form 17 March 2015 Accepted 9 May 2015
If an obstacle suddenly appears during walking, either the crossing step can be lengthened or the precrossing step shortened to avoid the obstacle. We investigated the effects of temporal constraints on dynamic stability during step adjustments. Twelve healthy young adults avoided a virtual white planar obstacle by lengthening or shortening their steps under free or constrained conditions. When constrained, participants had only one step to avoid the obstacle. The results indicated that center of mass (COM) displacement in the mediolateral (ML) direction and the COM velocity toward the swing-leg side during the crossing step were significantly increased in the long-constraint compared with the longfree condition. Consequently, the extrapolated COM (XcoM) position at the swing foot contact was also located further toward the swing-leg side. However, the distances between the XcoM and base of support (BOS) at the swing foot contact in the ML direction was unchanged because of greater lateral foot placement. In the anteriorposterior (AP) direction, temporal constraints led to greater AP COM displacement. The XcoM–BOS distance in the AP direction was unchanged in the long-constraint condition because of greater step length. However, the value became negative in the short-constraint condition, violating the conditions for dynamic stability, because step length adjustments were obstructed by the spatial constraints of the obstacles. These results suggest that temporal constraints affect postural stability in the AP and ML directions during step adjustments. AP and ML stability at swing foot contact are maintained through adjustments of step length and lateral foot placement, respectively. ß 2015 Elsevier B.V. All rights reserved.
Keywords: Obstacle negotiation Step adjustment Temporal constraints Lateral stability Dynamic stability
1. Introduction Falls among older individuals often occur during walking, and tripping and slipping are major contributors [1,2]. A previous systematic review reported that older adults are at greater risk of contacting obstacles under time-constrained conditions [3]. Most falls in older adults occur laterally because of deficits in lateral stability control [4,5]. Therefore, it is important to understand dynamic stability control in the mediolateral (ML) direction during obstacle avoidance under temporal constraints.
* Corresponding author at: Department of Physical Therapy, Tsukuba Internal University, 6-8-33 Manabe, Tsuchiura, Ibaraki 300-0051, Japan. Tel.: +81 29 826 6622; fax: +81 29 826 6776. E-mail addresses: [email protected] (W. Nakano), [email protected] (T. Fukaya), [email protected] (Y. Kanai), [email protected] (K. Akizuki), [email protected] (Y. Ohashi). http://dx.doi.org/10.1016/j.gaitpost.2015.05.004 0966-6362/ß 2015 Elsevier B.V. All rights reserved.
If an obstacle suddenly appears during walking, a long step strategy (LSS) or a short step strategy (SSS) can be used [6]. In an LSS, the obstacle is crossed using a lengthened crossing step. In an SSS, the precrossing step is shortened and the obstacle is crossed on the next step. Previous research has shown that sudden lengthening of the crossing step results in a loss of balance in the lateral direction [7]. Another study reported that sudden shortening of the precrossing step leads to falling without contacting an obstacle [6]. These results suggest that step adjustments under temporal constraints cause instability. To our knowledge, only one study has quantified dynamic stability during step adjustments under temporal constraints [8]. This study reported that stability in the anteroposterior (AP) direction deteriorated in an SSS, whereas ML stability was not affected by either an LSS or an SSS. In contrast, recent studies have shown that ML stability control is influenced by temporal constraints during volitional stepping and gait initiation [9– 12]. These discrepancies might be explained by the modulation of
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
2.3. Procedure
lateral foot placement. Caderby et al. [12] reported that temporal constraints led to greater center of mass (COM) shift toward the swing-leg side during gait initiation, whereas the margin of dynamic stability (MDS; the distance between the base of support and the extrapolated COM) at the swing foot contact in the ML direction was maintained because of regulation of lateral foot placement. Thus, the effects of temporal constraints on dynamic stability control during step adjustments should be evaluated by not only MDS but also COM motion and foot placement. The purpose of this study is to clarify the effects of temporal constraints on dynamic stability control during step adjustments. We presumed that AP stability control is affected by the SSS, whereas ML stability control is affected by the LSS. Our hypotheses were: (1) sudden lengthening of the crossing step leads to greater ML COM motion; (2) MDS in the ML direction is maintained by adjustments of lateral foot placement; and (3) MDS in the AP direction is affected by sudden shortening of the precrossing step.
Twenty-two reflective markers were attached to the participants’ bodies, positioned at: vertex, upper margin of sternum, left/ right middle of tragus, acromion, elbow joint, wrist joint, third metacarpophalangeal joint, greater trochanter, knee joint, lateral malleolus, and calcaneal tuberosity. Two reflective markers were also attached to the monitor to define the obstacle position. All participants stood in a natural upright posture at the starting position, and started walking with their left foot at a self-paced speed. The starting position was adjusted so that the middle of the right foot on the fourth step was located in the center of the obstacle (Fig. 1). The switch was located at the position of the third left heel contact. Tape showing the normal landing spot of the third left heel contact was placed on the switch. Participants were asked to walk several times to confirm whether the third left heel contact was on the switch and that the right foot on the fourth step was at the center of the obstacle without step adjustments. Gaze direction during the initial posture and the walk was not controlled. For obstacle conditions, participants avoided the obstacle by either lengthening or shortening their steps. The adjustment strategy to be performed was indicated using a arrow on the monitor (Fig. 1). Participants performed obstacle avoidance tasks under two temporal conditions: free or constrained. For the free condition, the obstacle and arrow were projected when participants stood at the starting position. For the constrained condition, the obstacle and arrow were projected when they stepped on the switch. Four obstacle conditions were collected in total: long-free (LF), long-constrained (LC), short-free (SF) and short-constrained (SC). Each participant performed five trials per condition. Participants could anticipate the obstacle appearance if the obstacle was invisible from the starting position (constrained conditions). To minimize anticipation, 10 walk-through trials were collected. The probability of obstacle appearance was 50% if the obstacle was invisible from the starting position. Also, participants were asked to step on the tape on the switch for both walk-through and constrained conditions. Participants initially performed three normal-walking trials with the obstacle and arrow not shown. Participants knew that the obstacle would not appear, and adjusted their step length without constraints to step on the switch or tape. Then, they performed at least two familiarization trials for each obstacle condition and walk-through condition followed by 20 obstacle trials and 10 walk-through trials. Trials were completely randomized. The main difference between the normal-walking and walk-through
2. Method 2.1. Participants Twelve healthy young adults volunteered for this study (six female; mean age 25.6 4.6 years; mean height 162.0 7.2 cm; mean weight 56.2 8.4 kg). All participants completed informed consent procedures approved by the local ethics committee. 2.2. Apparatus The present protocol replicated the protocol of Moraes et al. [8]. A liquid crystal display monitor (41.8 cm 55.6 cm; IIYAMA, ProLiteE-2473HDS-B, Tokyo, Japan) was embedded in a walkway (5.4 m 0.9 m). A piece of tempered glass was placed over the monitor so that participants could step over it normally. A virtual white planar obstacle measuring 10 cm (depth) 39 cm (length) was projected onto the middle of the monitor. The size of the obstacle was identical for all participants. A mat switch (operating force > 60 N; OJIDEN, OM-PVC623, Osaka, Japan) was connected to the monitor through the computer, and the obstacle was projected onto the monitor when participants stepped on the switch. Kinematic data were measured using an 8-camera motion analysis system (Oxford Metrics Group, Vicon Nexus, Oxford, UK) with a 100-Hz sampling rate.
Step1 z
y
1st foot contact
159
Step2
2nd foot contact
Step3
3rd foot contact
Step4
4th foot contact
5th foot contact
x
Mat switch
Direction of travel
Short
Long
Fig. 1. Schematic of the experimental set-up. Experimental set-up showing virtual white planar obstacle (10 cm 39 cm) and a arrow projected on the monitor. The arrow indicated the adjustment strategy (long or short) to be performed. The obstacle and the arrow were visible from the starting position for the free condition, and in the constrained condition they appeared when participants stepped on the mat switch.
160
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
conditions was probability of obstacle appearance (50% for walkthrough condition; 0% for normal-walking condition). Thus, the walk-through condition was more cognitively demanding than the normal-walking condition. These two conditions were compared to estimate the extents of effects of cognitive load on COM motion. 2.4. Data analysis Marker coordinates were filtered using a Butterworth low-pass filter with a cut-off frequency of 6 Hz. COM was calculated using a 14-segment model, comprising a single head and trunk and two upper arm, forearm, hand, thigh, shank and foot segments [13]. Foot contact was estimated using a stick picture and defined as the lowest point on the vertical axis of the heel or toe marker. The coordinates of the heel, toe and two markers on the obstacle were used to identify unsuccessful trials. If participants stepped on the obstacle or avoided it using the wrong strategy, the corresponding trials were excluded from the analysis. Because we aimed to quantify the effects of temporal constraints on dynamic stability during step adjustments, we measured spatiotemporal and stability parameters during Step3 (from the third to fourth swing foot contact) (Fig. 1). Step length was calculated as the AP distance between the heel markers at the time of swing foot contact. Step duration was calculated from the frame number. The range of motion of the COM (COM-ROM) was calculated from the maximum minus the minimum value in the AP and ML directions. The peak COM velocity (COM-velocity) was calculated from the maximum velocity toward the anterior and swing-leg side. Based on Hof et al. [14], extrapolated COM (XcoM) was calculated as:
XcoM ¼ COM þ
˙ COM
v0
3. Results 3.1. Success rates No fails were recorded in the LF and SF conditions. Eight out of the 60 total trials (each of the 12 participants performed 5 trials each) were failed in each of the LC and SC conditions. 3.2. Comparison of step spatiotemporal measures and COM motion between normal-walking and walk-through conditions There were no significant differences in step length, ML COMROM, or anterior COM-velocity. There were significant increases in step duration, AP COM-ROM and lateral COM-velocity in the walkthrough compared with the normal-walk condition. 3.3. Influence of temporal constraints on step spatiotemporal and postural parameters
;
˙ is COM velocity. v0 is where COM is actual COM position, and COM defined as:
v0 ¼
walk-through conditions were made using paired t-tests. The effects of temporal constraints (free and constrained) and step adjustment strategy (step lengthening and shortening) on each variable were assessed using two-way repeated-measures analysis of variance. For a significant interaction, post hoc tests (Shaffer’s modified sequentially rejective Bonferroni procedure) were used. Because we aimed to clarify the effects of temporal constraints, only the main effects of temporal constraints and simple main effect of temporal constraints at step lengthening (LF vs. LC) and shortening (SF vs. SC) were reported. Pearson’s correlation coefficient was used to determine the relationship between the variables. Statistical analysis was done using R statistical software 3.1.0. The level of statistical significance was set at a = 0.05.
rffiffiffi g ; l
where g = 9.81 m/s2 is acceleration due to gravity, and l is the length of the inverted pendulum. In this study, COM height was used as a substitute for l. MDS was defined as: MDS = BOS XcoM, where BOS is the base of support boundary. MDS was calculated at fourth swing foot contact (Fig. 4). MDS in the ML direction (MDS-ML) was calculated as the lateral distance between the XcoM and the lateral malleolus marker of the right foot. MDS in the AP direction (MDS-AP) was calculated as the AP distance between the XcoM and the toe marker of the right foot. A negative MDS indicates that the XcoM is located outside the BOS and is therefore unstable, and vice versa for a positive value. The XcoM shift toward the swing-leg side was calculated from the lateral distance between XcoM positions at the second and fourth swing foot contact times. Lateral foot placement was calculated from lateral distance between the heel markers at the time of the second and fourth swing foot contacts (Fig. 5). 2.5. Statistical analysis Mean values and standard deviations of dependent variables were calculated from the normal-walking and walk-through conditions, and successful trials for the obstacle conditions. For the walk-through condition, three trials were randomly extracted from 10 trials. Comparisons between the normal-walking and
For the step spatiotemporal parameters and COM motion along the progression axis, there was a significant interaction for both step length and step duration (Fig. 2). Post hoc tests indicated that step length in the LC condition was greater than in the LF condition, and that in the SC condition was shorter than in the SF condition. Step durations in the LC and SC conditions were longer than in the LF and SF conditions, respectively. A significant main effect of temporal constraint was found in AP COM-ROM. For anterior COM-velocity, a significant interaction was detected. Post hoc tests indicated that anterior COM-velocity in the LC condition was slower than in the LF condition, whereas no significant difference was found between the SF and SC conditions. Regarding COM motion in the ML direction, there was a significant interaction for both ML COM-ROM and lateral COMvelocity. As Fig. 3 reveals, ML COM-ROM in the LC and SC conditions was greater than in the LF and SF conditions, respectively. Lateral COM-velocity in the LC condition was faster than in the LF condition. Pearson’s correlation coefficient revealed significant positive correlations between step duration and ML COM-ROM (r = 0.86, p < 0.01) and lateral COM-velocity (r = 0.69, p < 0.01). The ANOVA revealed that no significant difference was found in MDS-ML (Fig. 4). However, MDS-AP in the SC condition was significantly reduced compared with the SF condition. A significant interaction was found for both XcoM shift and lateral foot placement (Fig. 5). Post hoc tests indicated that both XcoM shift and lateral foot placement in the LC condition reached a significantly higher value compared with the LF condition. Pearson’s correlation coefficient showed that there was significant positive correlation between XcoM shift and lateral foot placement (r = 0.85, p < 0.01).
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
(B)
**
** Step duration (s)
0.8 0.6 0.4 0.2 0
LF
LC SF Condition
**
1.0 0.6 0.4 0.2 0
SC
(C)
**
0.8
LF
LC SF Condition
SC
LC SF Condition
SC
(D)
¶¶
¶¶
1.0 0.8 0.6 0.4 0.2 0
LF
LC SF Condition
SC
Anterior COM-velocity (m/s)
Step length (m)
1.0
AP COM-ROM (m)
161
*
1.6 1.2 0.8 0.4 0
LF
Fig. 2. Comparison of step spatiotemporal parameters and COM motion along progression axis across obstacle conditions. (A) Comparison of step length across conditions. (B) Comparison of step duration across conditions. (C) Comparison of AP COM-ROM across conditions. (D) Comparison of anterior COM-velocity across conditions. Only statistics related to the effect of temporal constraints are reported in (A)–(D). COM, AP: center of mass and anteriorposterior, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. *Significant difference with p < 0.05. **Significant difference with p < 0.01. ôô Significant main effect of temporal constraints with p < 0.01.
4. Discussion Our results showed that ML COM-ROM and lateral COMvelocity in the LC condition were significantly increased compared
with the LF condition. One possible explanation for why ML COM motion increased under temporal constraints is changes in step spatiotemporal parameters along the progression axis. Previous research has reported that stepping over an obstacle under
(A) swing-leg side
0.25
0.25
Trajectory of COM (m)
Linear velocity
0.15
0.15
COM-velocity
0.10
0.10
Trajectory
0.05
0.05
0
0
-0.05
-0.05
-0.10
-0.10 -0.15
-0.15 COM-ROM
-0.20
-0.20
stance-leg side
-0.25
-0.25 Step1
Step2
Step3
(B) 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
Step
(C)
**
LF
**
LC SF Condition
SC
Lateral COM-velocity (m/s)
ML COM-ROM (m)
Linear velocity of COM (m/s)
0.20
0.20
**
0.25 0.2 0.15 0.1 0.05 0
LF
LC SF Condition
SC
Fig. 3. Comparison of motion of COM in mediolateral direction across obstacle conditions. (A) Example of motion of COM in ML direction from one representative participant in one trial for the LC condition. ML COM-ROM and ML COM-velocity were calculated during Step3 (from third to fourth swing foot contact). (B) Comparison of ML COM-ROM across conditions. (C) Comparison of lateral COM-velocity across conditions. Only statistics related to the effect of temporal constraints are reported in (B) and (C). COM, ML: center of mass and mediolateral, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. **Significant difference with p < 0.01.
162
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
(B)
XcoM position
Toe marker MDS-AP
4th foot contact
Ankle marker XcoM position
MDS-ML
Actual COM position
Actual COM position
3rd foot contact
(D)
(C) 0.08
MDS-AP (m)
MDS-ML (m)
0.1 0.06 0.04 0.02 0
LF
LC SF Condition
SC
0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2
**
LF
LC SF Condition
SC
Fig. 4. Comparison of MDS-ML and MDS-AP across obstacle conditions. (A) Example of MDS-ML. MDS-ML was calculated as the lateral distance between the XcoM and the ankle marker (lateral malleolus). (B) Example of MDS-AP. MDS-AP was calculated as the AP distance between the XcoM and the toe marker. MDS-ML and MDS-AP were calculated at the fourth swing foot contact. A positive MDS indicates that the XcoM is located within the base of support. A negative MDS indicates XcoM is located outside the base of support and therefore indicates instability. (C) Comparison of MDS-ML across conditions. (D) Comparison of MDS-AP across conditions. Only statistics related to the effect of temporal constraints are reported in (D). MDS, COM, XcoM, AP, ML: margin of dynamic stability, center of mass, extrapolated center of mass, anteriorposterior and mediolateral, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. **Significant difference with p < 0.01.
temporal constraints increases both crossing step length and step duration [6]. Prolonging swing duration results in increased lateral COM motion [15,16]. Supporting these studies, our results showed that both step length and step duration were increased in the LC condition compared with the LF condition. Moreover, significant positive correlations were found between step duration and ML COM-ROM and lateral COM-velocity, i.e., the greater the step duration, the greater and faster the lateral COM motion. Stepping over an obstacle under no temporal constraints does not appear to increase ML COM motion despite increased step duration [17], because the COM position is kept close to the center of pressure (COP) position [18]. For volitional stepping, ML anticipatory postural adjustments (APA) minimize ML COM motion [19–21]. It is thought that predictive control of ML COM motion during obstacle negotiation is analogous to APA control during volitional stepping [15]. Under temporal constraints, ML APA is modulated as a reactive control mode to minimize ML COM motion during rapid leg flexion [11]. The present study demonstrates that temporal constraints leads to greater ML COM motion during obstacle avoidance, whereas whether ML COM motion is minimized by reactive corrections is still not understood. Measuring COM and its coordination with COP would provide a more complete understanding of the effects of temporal constraints on lateral stability control. Many studies report that postural control is cognitively demanding [22]. In our study, participants knew that the obstacle might appear at Step3 in some trials if the obstacle was invisible from the starting position (walk-through and constrained conditions). Thus, these conditions were more cognitively demanding than the normal-walking and free conditions. In fact, significant differences in COM kinematics were found between the normalwalking and walk-through conditions, suggesting that cognitive
load may in part cause greater ML COM motion in the LC condition. However, there was no significant difference in ML COM-ROM between the normal-walking and walk-through conditions, whereas there was a 53% increase in the LC compared with the LF condition. Also, there was no significant difference in lateral COM-velocity between the SF and SC conditions, whereas there was a 46% increase in the LC compared with the LF condition. These results cannot be explained by cognitive load. Thus, ML COM motion in the LC condition may have been caused by temporal constraints rather than by cognitive load. As a consequence of greater ML COM motion, the XcoM position at the swing foot contact in the LC condition was also located further toward the swing-leg side compared with in the LF condition. However, because the swing foot location also shifts toward the swing-leg side, MDS-ML does not change. Previous research reports that MDS-ML is not affected by step adjustments under temporal constraints [8]. However, it should be emphasized that MDS-ML remains constant because of lateral step adjustments. This notion is further supported by strong positive correlation between XcoM shift and lateral foot placement (r = 0.85), i.e., the greater the XcoM shift toward the swing-leg side, the greater the lateral step adjustments. Caderby et al. [12] reported strong positive correlation between XcoM position and step width during gait initiation. These results support the idea that XcoM position may function as a balance control parameter [12,23–26]. AP stability during obstacle negotiation is maintained by increasing forward step distance and decelerating forward COM velocity [15]. Consistent with this view, step length increased (average 6.7 cm) and anterior COM-velocity decreased in the LC condition compared with the LF condition. Consequently, although
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
Heel marker
163
XcoM position
XcoM shift
2nd foot contact
Lateral foot placement
(B) *
**
(D)
XcoM shift (m)
0.08 0.06 0.04 0.02 0 -0.02 -0.04 -0.06
4th foot contact
0.08 0.06
LC
SF
SC
Lateral foot placement (m)
**
0.06 0.04 0.02 0
XcoM shift (m)
0.04 LF
(C)
-0.02 -0.04 -0.06
r=0.85, p
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Effects of temporal constraints on medio-lateral stability when negotiating obstacles Wataru Nakano a,*, Takashi Fukaya a, Yoshihide Kanai b, Kazunori Akizuki c, Yukari Ohashi b a
Department of Physical Therapy, Tsukuba International University, 6-8-33 Manabe, Tsuchiura, Ibaraki, Japan Department of Physical Therapy, Ibaraki Prefectural University of Health Sciences, 4669-2 Ami, Ami, Ibaraki, Japan c Department of Physical Therapy, Mejiro University, 320 Ukiya, Iwatsuki, Saitama, Japan b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 28 July 2014 Received in revised form 17 March 2015 Accepted 9 May 2015
If an obstacle suddenly appears during walking, either the crossing step can be lengthened or the precrossing step shortened to avoid the obstacle. We investigated the effects of temporal constraints on dynamic stability during step adjustments. Twelve healthy young adults avoided a virtual white planar obstacle by lengthening or shortening their steps under free or constrained conditions. When constrained, participants had only one step to avoid the obstacle. The results indicated that center of mass (COM) displacement in the mediolateral (ML) direction and the COM velocity toward the swing-leg side during the crossing step were significantly increased in the long-constraint compared with the longfree condition. Consequently, the extrapolated COM (XcoM) position at the swing foot contact was also located further toward the swing-leg side. However, the distances between the XcoM and base of support (BOS) at the swing foot contact in the ML direction was unchanged because of greater lateral foot placement. In the anteriorposterior (AP) direction, temporal constraints led to greater AP COM displacement. The XcoM–BOS distance in the AP direction was unchanged in the long-constraint condition because of greater step length. However, the value became negative in the short-constraint condition, violating the conditions for dynamic stability, because step length adjustments were obstructed by the spatial constraints of the obstacles. These results suggest that temporal constraints affect postural stability in the AP and ML directions during step adjustments. AP and ML stability at swing foot contact are maintained through adjustments of step length and lateral foot placement, respectively. ß 2015 Elsevier B.V. All rights reserved.
Keywords: Obstacle negotiation Step adjustment Temporal constraints Lateral stability Dynamic stability
1. Introduction Falls among older individuals often occur during walking, and tripping and slipping are major contributors [1,2]. A previous systematic review reported that older adults are at greater risk of contacting obstacles under time-constrained conditions [3]. Most falls in older adults occur laterally because of deficits in lateral stability control [4,5]. Therefore, it is important to understand dynamic stability control in the mediolateral (ML) direction during obstacle avoidance under temporal constraints.
* Corresponding author at: Department of Physical Therapy, Tsukuba Internal University, 6-8-33 Manabe, Tsuchiura, Ibaraki 300-0051, Japan. Tel.: +81 29 826 6622; fax: +81 29 826 6776. E-mail addresses: [email protected] (W. Nakano), [email protected] (T. Fukaya), [email protected] (Y. Kanai), [email protected] (K. Akizuki), [email protected] (Y. Ohashi). http://dx.doi.org/10.1016/j.gaitpost.2015.05.004 0966-6362/ß 2015 Elsevier B.V. All rights reserved.
If an obstacle suddenly appears during walking, a long step strategy (LSS) or a short step strategy (SSS) can be used [6]. In an LSS, the obstacle is crossed using a lengthened crossing step. In an SSS, the precrossing step is shortened and the obstacle is crossed on the next step. Previous research has shown that sudden lengthening of the crossing step results in a loss of balance in the lateral direction [7]. Another study reported that sudden shortening of the precrossing step leads to falling without contacting an obstacle [6]. These results suggest that step adjustments under temporal constraints cause instability. To our knowledge, only one study has quantified dynamic stability during step adjustments under temporal constraints [8]. This study reported that stability in the anteroposterior (AP) direction deteriorated in an SSS, whereas ML stability was not affected by either an LSS or an SSS. In contrast, recent studies have shown that ML stability control is influenced by temporal constraints during volitional stepping and gait initiation [9– 12]. These discrepancies might be explained by the modulation of
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
2.3. Procedure
lateral foot placement. Caderby et al. [12] reported that temporal constraints led to greater center of mass (COM) shift toward the swing-leg side during gait initiation, whereas the margin of dynamic stability (MDS; the distance between the base of support and the extrapolated COM) at the swing foot contact in the ML direction was maintained because of regulation of lateral foot placement. Thus, the effects of temporal constraints on dynamic stability control during step adjustments should be evaluated by not only MDS but also COM motion and foot placement. The purpose of this study is to clarify the effects of temporal constraints on dynamic stability control during step adjustments. We presumed that AP stability control is affected by the SSS, whereas ML stability control is affected by the LSS. Our hypotheses were: (1) sudden lengthening of the crossing step leads to greater ML COM motion; (2) MDS in the ML direction is maintained by adjustments of lateral foot placement; and (3) MDS in the AP direction is affected by sudden shortening of the precrossing step.
Twenty-two reflective markers were attached to the participants’ bodies, positioned at: vertex, upper margin of sternum, left/ right middle of tragus, acromion, elbow joint, wrist joint, third metacarpophalangeal joint, greater trochanter, knee joint, lateral malleolus, and calcaneal tuberosity. Two reflective markers were also attached to the monitor to define the obstacle position. All participants stood in a natural upright posture at the starting position, and started walking with their left foot at a self-paced speed. The starting position was adjusted so that the middle of the right foot on the fourth step was located in the center of the obstacle (Fig. 1). The switch was located at the position of the third left heel contact. Tape showing the normal landing spot of the third left heel contact was placed on the switch. Participants were asked to walk several times to confirm whether the third left heel contact was on the switch and that the right foot on the fourth step was at the center of the obstacle without step adjustments. Gaze direction during the initial posture and the walk was not controlled. For obstacle conditions, participants avoided the obstacle by either lengthening or shortening their steps. The adjustment strategy to be performed was indicated using a arrow on the monitor (Fig. 1). Participants performed obstacle avoidance tasks under two temporal conditions: free or constrained. For the free condition, the obstacle and arrow were projected when participants stood at the starting position. For the constrained condition, the obstacle and arrow were projected when they stepped on the switch. Four obstacle conditions were collected in total: long-free (LF), long-constrained (LC), short-free (SF) and short-constrained (SC). Each participant performed five trials per condition. Participants could anticipate the obstacle appearance if the obstacle was invisible from the starting position (constrained conditions). To minimize anticipation, 10 walk-through trials were collected. The probability of obstacle appearance was 50% if the obstacle was invisible from the starting position. Also, participants were asked to step on the tape on the switch for both walk-through and constrained conditions. Participants initially performed three normal-walking trials with the obstacle and arrow not shown. Participants knew that the obstacle would not appear, and adjusted their step length without constraints to step on the switch or tape. Then, they performed at least two familiarization trials for each obstacle condition and walk-through condition followed by 20 obstacle trials and 10 walk-through trials. Trials were completely randomized. The main difference between the normal-walking and walk-through
2. Method 2.1. Participants Twelve healthy young adults volunteered for this study (six female; mean age 25.6 4.6 years; mean height 162.0 7.2 cm; mean weight 56.2 8.4 kg). All participants completed informed consent procedures approved by the local ethics committee. 2.2. Apparatus The present protocol replicated the protocol of Moraes et al. [8]. A liquid crystal display monitor (41.8 cm 55.6 cm; IIYAMA, ProLiteE-2473HDS-B, Tokyo, Japan) was embedded in a walkway (5.4 m 0.9 m). A piece of tempered glass was placed over the monitor so that participants could step over it normally. A virtual white planar obstacle measuring 10 cm (depth) 39 cm (length) was projected onto the middle of the monitor. The size of the obstacle was identical for all participants. A mat switch (operating force > 60 N; OJIDEN, OM-PVC623, Osaka, Japan) was connected to the monitor through the computer, and the obstacle was projected onto the monitor when participants stepped on the switch. Kinematic data were measured using an 8-camera motion analysis system (Oxford Metrics Group, Vicon Nexus, Oxford, UK) with a 100-Hz sampling rate.
Step1 z
y
1st foot contact
159
Step2
2nd foot contact
Step3
3rd foot contact
Step4
4th foot contact
5th foot contact
x
Mat switch
Direction of travel
Short
Long
Fig. 1. Schematic of the experimental set-up. Experimental set-up showing virtual white planar obstacle (10 cm 39 cm) and a arrow projected on the monitor. The arrow indicated the adjustment strategy (long or short) to be performed. The obstacle and the arrow were visible from the starting position for the free condition, and in the constrained condition they appeared when participants stepped on the mat switch.
160
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
conditions was probability of obstacle appearance (50% for walkthrough condition; 0% for normal-walking condition). Thus, the walk-through condition was more cognitively demanding than the normal-walking condition. These two conditions were compared to estimate the extents of effects of cognitive load on COM motion. 2.4. Data analysis Marker coordinates were filtered using a Butterworth low-pass filter with a cut-off frequency of 6 Hz. COM was calculated using a 14-segment model, comprising a single head and trunk and two upper arm, forearm, hand, thigh, shank and foot segments [13]. Foot contact was estimated using a stick picture and defined as the lowest point on the vertical axis of the heel or toe marker. The coordinates of the heel, toe and two markers on the obstacle were used to identify unsuccessful trials. If participants stepped on the obstacle or avoided it using the wrong strategy, the corresponding trials were excluded from the analysis. Because we aimed to quantify the effects of temporal constraints on dynamic stability during step adjustments, we measured spatiotemporal and stability parameters during Step3 (from the third to fourth swing foot contact) (Fig. 1). Step length was calculated as the AP distance between the heel markers at the time of swing foot contact. Step duration was calculated from the frame number. The range of motion of the COM (COM-ROM) was calculated from the maximum minus the minimum value in the AP and ML directions. The peak COM velocity (COM-velocity) was calculated from the maximum velocity toward the anterior and swing-leg side. Based on Hof et al. [14], extrapolated COM (XcoM) was calculated as:
XcoM ¼ COM þ
˙ COM
v0
3. Results 3.1. Success rates No fails were recorded in the LF and SF conditions. Eight out of the 60 total trials (each of the 12 participants performed 5 trials each) were failed in each of the LC and SC conditions. 3.2. Comparison of step spatiotemporal measures and COM motion between normal-walking and walk-through conditions There were no significant differences in step length, ML COMROM, or anterior COM-velocity. There were significant increases in step duration, AP COM-ROM and lateral COM-velocity in the walkthrough compared with the normal-walk condition. 3.3. Influence of temporal constraints on step spatiotemporal and postural parameters
;
˙ is COM velocity. v0 is where COM is actual COM position, and COM defined as:
v0 ¼
walk-through conditions were made using paired t-tests. The effects of temporal constraints (free and constrained) and step adjustment strategy (step lengthening and shortening) on each variable were assessed using two-way repeated-measures analysis of variance. For a significant interaction, post hoc tests (Shaffer’s modified sequentially rejective Bonferroni procedure) were used. Because we aimed to clarify the effects of temporal constraints, only the main effects of temporal constraints and simple main effect of temporal constraints at step lengthening (LF vs. LC) and shortening (SF vs. SC) were reported. Pearson’s correlation coefficient was used to determine the relationship between the variables. Statistical analysis was done using R statistical software 3.1.0. The level of statistical significance was set at a = 0.05.
rffiffiffi g ; l
where g = 9.81 m/s2 is acceleration due to gravity, and l is the length of the inverted pendulum. In this study, COM height was used as a substitute for l. MDS was defined as: MDS = BOS XcoM, where BOS is the base of support boundary. MDS was calculated at fourth swing foot contact (Fig. 4). MDS in the ML direction (MDS-ML) was calculated as the lateral distance between the XcoM and the lateral malleolus marker of the right foot. MDS in the AP direction (MDS-AP) was calculated as the AP distance between the XcoM and the toe marker of the right foot. A negative MDS indicates that the XcoM is located outside the BOS and is therefore unstable, and vice versa for a positive value. The XcoM shift toward the swing-leg side was calculated from the lateral distance between XcoM positions at the second and fourth swing foot contact times. Lateral foot placement was calculated from lateral distance between the heel markers at the time of the second and fourth swing foot contacts (Fig. 5). 2.5. Statistical analysis Mean values and standard deviations of dependent variables were calculated from the normal-walking and walk-through conditions, and successful trials for the obstacle conditions. For the walk-through condition, three trials were randomly extracted from 10 trials. Comparisons between the normal-walking and
For the step spatiotemporal parameters and COM motion along the progression axis, there was a significant interaction for both step length and step duration (Fig. 2). Post hoc tests indicated that step length in the LC condition was greater than in the LF condition, and that in the SC condition was shorter than in the SF condition. Step durations in the LC and SC conditions were longer than in the LF and SF conditions, respectively. A significant main effect of temporal constraint was found in AP COM-ROM. For anterior COM-velocity, a significant interaction was detected. Post hoc tests indicated that anterior COM-velocity in the LC condition was slower than in the LF condition, whereas no significant difference was found between the SF and SC conditions. Regarding COM motion in the ML direction, there was a significant interaction for both ML COM-ROM and lateral COMvelocity. As Fig. 3 reveals, ML COM-ROM in the LC and SC conditions was greater than in the LF and SF conditions, respectively. Lateral COM-velocity in the LC condition was faster than in the LF condition. Pearson’s correlation coefficient revealed significant positive correlations between step duration and ML COM-ROM (r = 0.86, p < 0.01) and lateral COM-velocity (r = 0.69, p < 0.01). The ANOVA revealed that no significant difference was found in MDS-ML (Fig. 4). However, MDS-AP in the SC condition was significantly reduced compared with the SF condition. A significant interaction was found for both XcoM shift and lateral foot placement (Fig. 5). Post hoc tests indicated that both XcoM shift and lateral foot placement in the LC condition reached a significantly higher value compared with the LF condition. Pearson’s correlation coefficient showed that there was significant positive correlation between XcoM shift and lateral foot placement (r = 0.85, p < 0.01).
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
(B)
**
** Step duration (s)
0.8 0.6 0.4 0.2 0
LF
LC SF Condition
**
1.0 0.6 0.4 0.2 0
SC
(C)
**
0.8
LF
LC SF Condition
SC
LC SF Condition
SC
(D)
¶¶
¶¶
1.0 0.8 0.6 0.4 0.2 0
LF
LC SF Condition
SC
Anterior COM-velocity (m/s)
Step length (m)
1.0
AP COM-ROM (m)
161
*
1.6 1.2 0.8 0.4 0
LF
Fig. 2. Comparison of step spatiotemporal parameters and COM motion along progression axis across obstacle conditions. (A) Comparison of step length across conditions. (B) Comparison of step duration across conditions. (C) Comparison of AP COM-ROM across conditions. (D) Comparison of anterior COM-velocity across conditions. Only statistics related to the effect of temporal constraints are reported in (A)–(D). COM, AP: center of mass and anteriorposterior, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. *Significant difference with p < 0.05. **Significant difference with p < 0.01. ôô Significant main effect of temporal constraints with p < 0.01.
4. Discussion Our results showed that ML COM-ROM and lateral COMvelocity in the LC condition were significantly increased compared
with the LF condition. One possible explanation for why ML COM motion increased under temporal constraints is changes in step spatiotemporal parameters along the progression axis. Previous research has reported that stepping over an obstacle under
(A) swing-leg side
0.25
0.25
Trajectory of COM (m)
Linear velocity
0.15
0.15
COM-velocity
0.10
0.10
Trajectory
0.05
0.05
0
0
-0.05
-0.05
-0.10
-0.10 -0.15
-0.15 COM-ROM
-0.20
-0.20
stance-leg side
-0.25
-0.25 Step1
Step2
Step3
(B) 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0
Step
(C)
**
LF
**
LC SF Condition
SC
Lateral COM-velocity (m/s)
ML COM-ROM (m)
Linear velocity of COM (m/s)
0.20
0.20
**
0.25 0.2 0.15 0.1 0.05 0
LF
LC SF Condition
SC
Fig. 3. Comparison of motion of COM in mediolateral direction across obstacle conditions. (A) Example of motion of COM in ML direction from one representative participant in one trial for the LC condition. ML COM-ROM and ML COM-velocity were calculated during Step3 (from third to fourth swing foot contact). (B) Comparison of ML COM-ROM across conditions. (C) Comparison of lateral COM-velocity across conditions. Only statistics related to the effect of temporal constraints are reported in (B) and (C). COM, ML: center of mass and mediolateral, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. **Significant difference with p < 0.01.
162
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
(B)
XcoM position
Toe marker MDS-AP
4th foot contact
Ankle marker XcoM position
MDS-ML
Actual COM position
Actual COM position
3rd foot contact
(D)
(C) 0.08
MDS-AP (m)
MDS-ML (m)
0.1 0.06 0.04 0.02 0
LF
LC SF Condition
SC
0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2
**
LF
LC SF Condition
SC
Fig. 4. Comparison of MDS-ML and MDS-AP across obstacle conditions. (A) Example of MDS-ML. MDS-ML was calculated as the lateral distance between the XcoM and the ankle marker (lateral malleolus). (B) Example of MDS-AP. MDS-AP was calculated as the AP distance between the XcoM and the toe marker. MDS-ML and MDS-AP were calculated at the fourth swing foot contact. A positive MDS indicates that the XcoM is located within the base of support. A negative MDS indicates XcoM is located outside the base of support and therefore indicates instability. (C) Comparison of MDS-ML across conditions. (D) Comparison of MDS-AP across conditions. Only statistics related to the effect of temporal constraints are reported in (D). MDS, COM, XcoM, AP, ML: margin of dynamic stability, center of mass, extrapolated center of mass, anteriorposterior and mediolateral, respectively. LF, LC, SF, SC: long-free, long-constrained, short-free and short-constrained, respectively. Reported values are means 1 standard deviation. **Significant difference with p < 0.01.
temporal constraints increases both crossing step length and step duration [6]. Prolonging swing duration results in increased lateral COM motion [15,16]. Supporting these studies, our results showed that both step length and step duration were increased in the LC condition compared with the LF condition. Moreover, significant positive correlations were found between step duration and ML COM-ROM and lateral COM-velocity, i.e., the greater the step duration, the greater and faster the lateral COM motion. Stepping over an obstacle under no temporal constraints does not appear to increase ML COM motion despite increased step duration [17], because the COM position is kept close to the center of pressure (COP) position [18]. For volitional stepping, ML anticipatory postural adjustments (APA) minimize ML COM motion [19–21]. It is thought that predictive control of ML COM motion during obstacle negotiation is analogous to APA control during volitional stepping [15]. Under temporal constraints, ML APA is modulated as a reactive control mode to minimize ML COM motion during rapid leg flexion [11]. The present study demonstrates that temporal constraints leads to greater ML COM motion during obstacle avoidance, whereas whether ML COM motion is minimized by reactive corrections is still not understood. Measuring COM and its coordination with COP would provide a more complete understanding of the effects of temporal constraints on lateral stability control. Many studies report that postural control is cognitively demanding [22]. In our study, participants knew that the obstacle might appear at Step3 in some trials if the obstacle was invisible from the starting position (walk-through and constrained conditions). Thus, these conditions were more cognitively demanding than the normal-walking and free conditions. In fact, significant differences in COM kinematics were found between the normalwalking and walk-through conditions, suggesting that cognitive
load may in part cause greater ML COM motion in the LC condition. However, there was no significant difference in ML COM-ROM between the normal-walking and walk-through conditions, whereas there was a 53% increase in the LC compared with the LF condition. Also, there was no significant difference in lateral COM-velocity between the SF and SC conditions, whereas there was a 46% increase in the LC compared with the LF condition. These results cannot be explained by cognitive load. Thus, ML COM motion in the LC condition may have been caused by temporal constraints rather than by cognitive load. As a consequence of greater ML COM motion, the XcoM position at the swing foot contact in the LC condition was also located further toward the swing-leg side compared with in the LF condition. However, because the swing foot location also shifts toward the swing-leg side, MDS-ML does not change. Previous research reports that MDS-ML is not affected by step adjustments under temporal constraints [8]. However, it should be emphasized that MDS-ML remains constant because of lateral step adjustments. This notion is further supported by strong positive correlation between XcoM shift and lateral foot placement (r = 0.85), i.e., the greater the XcoM shift toward the swing-leg side, the greater the lateral step adjustments. Caderby et al. [12] reported strong positive correlation between XcoM position and step width during gait initiation. These results support the idea that XcoM position may function as a balance control parameter [12,23–26]. AP stability during obstacle negotiation is maintained by increasing forward step distance and decelerating forward COM velocity [15]. Consistent with this view, step length increased (average 6.7 cm) and anterior COM-velocity decreased in the LC condition compared with the LF condition. Consequently, although
W. Nakano et al. / Gait & Posture 42 (2015) 158–164
(A)
Heel marker
163
XcoM position
XcoM shift
2nd foot contact
Lateral foot placement
(B) *
**
(D)
XcoM shift (m)
0.08 0.06 0.04 0.02 0 -0.02 -0.04 -0.06
4th foot contact
0.08 0.06
LC
SF
SC
Lateral foot placement (m)
**
0.06 0.04 0.02 0
XcoM shift (m)
0.04 LF
(C)
-0.02 -0.04 -0.06
r=0.85, p
